Protein structure information is indispensable for the design of effective drugs. Designers need detailed structures of proteins, to atomic resolution, so that they can tailor their drugs to interact with specific target areas in a protein molecule. Conventionally, x-ray crystallography has been used to elucidate the three-dimensional (3-D) structure of proteins. This technique can accurately identify the location of atoms by diffracting x-rays from innumerable protein molecules stacked up in an ordered form, so called “crystal”. However, crystallographers have yet to overcome the fact that most proteins do not readily form ordered assemblies. Particularly important but extremely difficult to crystallize are membrane proteins. Membrane proteins are not soluble, therefore conventional three-dimensional crystallization techniques usually do not work for them.
To date, the number of transmembrane proteins crystallized remains small. Indeed, recently it was noted that only 26 out of the approximately 1000 protein structure holdings in the Protein Data Bank (PDB) are membrane proteins (Nature Biotechnology 905 (2000)). In a 1998 report of the Committee for the National Magnetic Resonance Collaboration, it was stated that “even though membrane proteins represent 30% of the proteome, relatively little is known about the structure of these proteins, because of their resistance to crystallization.”
In general, membrane proteins are comprised of hydrophobic portions within their transmembrane regions, which render them insoluble in water. Consequently, unlike soluble proteins, membrane proteins do not form the monodispersed, isotropic solutions needed to grow crystals. This accounts for the near absence of structure information in PDB for transmembrane portions of most membrane proteins. In contrast, extracellular domains of membrane proteins which lack the hydrophobic regions, have been successfully crystallized.
A few techniques have been developed for membrane protein crystallization (Garavito, R. M. & Picot, D. Methods 1, 57-69 (1990); Kuelbrandt, W. Q. Rev. Biophysics 25, 1-49 (1992)). These techniques include i) application of conventional three-dimensional crystallization techniques directly to a preparation of detergent-solubilized membrane protein, e.g., by adding precipitation agents like ammonium sulfate or polyethylene glycol; and (ii) reconstitution of membrane proteins into lipid bilayers by detergent removal. In both of these methods, the crystallization process is hindered by the presence of detergent. For example, the detergent may inhibit crystal nucleation and growth. The presence of the detergent may require time-consuming and exhaustive detergent screening to determine whether the solubilized protein is functionally active. Furthermore, the crystals may include the detergent, therefore different detergents may yield different crystals of the same protein. Detergents also may effect protein orientation; i.e. alternating molecules face up or down or two layers stack up with an in plane axis of two-fold symmetry. Furthermore, the processes involving detergents are tedious and require long crystallization times, (e.g., typically several days to several weeks). In addition, these methods require large quantities (several 100 milligrams to grams) of purified protein that is not easily available in most cases. Also, designing crystallization experiments for a new system is not straight forward, as the same protocol does not always work for other systems.
Recently Landau and Rosenbusch demonstrated formation of three-dimensional crystals of bacteriorhodopsin by emulating the natural environment of the membrane protein in bicontinuous lipidic cubic phases (Landau, E. M. & Rosenbusch, J. P. Proc. Natl. Acad. Sci. USA 93, 14532-14535 (1996)). Their method included a protein delipidation process and required the use of detergents and precipitants. The crystals produced by this method were small, 20-40 μm in diameter and 5 μm thick, but diffracted x-rays from an intense microbeam source to 2.5 Å resolution. These harsh conditions may adversely affect the structure and function of more vulnerable membrane proteins. Furthermore, their method requires designing and building an artificial membrane with a different lattice size for each membrane protein. This does not appear to be an easy task.
Planar biological membranes are becoming of increasing interest because they provide a natural fluid membrane milieu that is critically important to the function of membrane proteins. This environment is ideal for immobilizing proteins under nondenaturing conditions and in a well-defined orientation. Two-dimensional crystallization of soluble proteins on lipid monolayers has been attempted for several systems (Komberg, R. D. & Darst, S. A. Curr. Opinion in Struct. Biol. 1, 642-646 (1991); Newman, R. Electron Microscopy Reviews (1991)). Formation of two-dimensional crystals non-soluble integral membrane proteins by compression of a monolayer at an air-water interface, has not yet been explored. Glaeser demonstrated preparation of thin, flat electron microscopy specimen from monolayers spread at the air-water interface of a monolayer trough, from native purple membranes including naturally occurring two-dimensional crystals of bacteriorhodopsin (Glaeser, R. A. Ann. Rev. Phys. Chem. 36, 243-75 (1985)).
In phospholipid systems, finite two-dimensional domains have been produced by monolayer compression at an air-water interface for several lipids. In studies concerning the compression of the lipid DPPC in monolayers, the presence of long-range orientational order throughout the solid domains produced at the coexistence region between two phases was found (Moy, V. T. et al. J. Phys. Chem. 92, 5233 (1988)). The experimental approach used in these studies does not provide concrete evidence on the crystalline order of the solid domains. However, the evidence of this long-range orientational order together with theoretical analysis of structure formation in lipid monolayers may provide a basis for understanding the production of structures in more complex mixed lipid-protein monolayer systems (McConnell, H. M. Annu. Rev. Phys. Chem. 42, 171-95 (1991)).